A recent source of concern for the security of neural networks is the\nemergence of clean-label dataset poisoning attacks, wherein correctly labeled\npoison samples are injected into the training dataset. While these poison\nsamples look legitimate to the human observer, they contain malicious\ncharacteristics that trigger a targeted misclassification during inference. We\npropose a scalable and transferable clean-label poisoning attack against\ntransfer learning, which creates poison images with their center close to the\ntarget image in the feature space. Our attack, Bullseye Polytope, improves the\nattack success rate of the current state-of-the-art by 26.75% in end-to-end\ntransfer learning, while increasing attack speed by a factor of 12. We further\nextend Bullseye Polytope to a more practical attack model by including multiple\nimages of the same object (e.g., from different angles) when crafting the\npoison samples. We demonstrate that this extension improves attack\ntransferability by over 16% to unseen images (of the same object) without using\nextra poison samples.\n